Size-dependent Photoconductivity in MBE-Grown GaN
Size-dependent Photoconductivity in
MBE-Grown GaN?Nanowires
Raffaella Calarco,*Michel Marso,Thomas Richter,Ali I.Aykanat,Ralph Meijers,
Andre′v.d.Hart,Toma Stoica,and Hans Lu1th
Institute of Thin Films and Interfaces(ISG1)and cni-Center of Nanoelectronic
Systems for Information Technology,Research Centre Ju¨lich,52425Ju¨lich,Germany
Received January6,2005;Revised Manuscript Received March15,2005
ABSTRACT
We report on electrical transport in the dark and under ultraviolet(UV)illumination through GaN nanowhiskers grown by molecular beam epitaxy(MBE),which is sensitively dependent on the column diameter.This new effect is quantitatively described by a size dependent surface recombination mechanism.The essential ingredient for the interpretation of this effect is a diameter dependent recombination barrier,which arises from the interplay between column diameter and space charge layer extension at the column surface.
The bottom-up approach to growing nanowires(whiskers) in several semiconductor material systems(Si,1III-V,2-6 III-nitride7-9)with the aim of quantum electronic,optoelec-tronic,and sensor applications has attracted considerable interest worldwide.Even though sophisticated device struc-tures such as RTDs,SETs,field effect transistors,biosensors, and even logic gates could already be realized by these self-organized nanowires,10-15many fundamental questions about the internal electronic structure,the effect of the large surface in comparison to its bulk,and size dependent transport phenomena remain unanswered up to now.
In this context our investigation of MBE-grown GaN nanowires demonstrates,for the first time to our knowledge, the effect of surface Fermi-level pinning16and its interplay with the nanowire dimensions on the recombination behavior of electron-hole pairs in photoconductivity through these wires.
Due to surface Fermi-level pinning within the forbidden band,GaN wires,as do most other semiconductor wires, exhibit a depletion space charge layer with an extension of the order of the wire diameter.Depending on wire thickness and doping,completely depleted wires or wires with thin conducting channels exist.Depending on the applied voltage, charge limited currents,characteristic for insulators,or ohmic behavior is observed.These effects are explicitly obvious under light-induced carrier excitation.In corresponding photoconductivity experiments,therefore,photoexcited elec-trons and holes are more or less spatially separated from each other depending on the extension of the surface space charge layer in comparison with the wire diameter.When surface recombination of nonequilibrium electron-hole pairs through surface traps is the prevailing mechanism,then the recombination rate and therefore also the absolute amount of the photocurrent density should depend very sensitively on the wire diameter.This is indeed found in the present investigations.
The GaN nanowires are grown by radio frequency plasma-assisted MBE on Si(111)substrates.The whisker density and diameter,20-500nm,are controlled by means of the III/V ratio.Below stoichiometry(defined as the III/V ratio at which the growth rate saturates)the epitaxial growth proceeds nominally under N-rich conditions,which leads to the desired columnar morphologies.The growth conditions are described elsewhere.17For a detailed experimental description of the growth mechanisms see the extensive work of Calleja et al.8 After epitaxial growth,the nanowires are released from the native Si(111)substrate by exposure to an ultrasonic bath and deposited on a Si(100)host substrate covered with an insulation layer of300nm SiO2.Ti(10nm)/Au(100nm) contacts patterned by electron beam lithography allow the electrical and optoelectrical characterization of the nano-wires(Figure1).To check for possible surface effects on the transport measurements due to atmosphere,the first investigations were performed under a vacuum of10-5mbar, but no difference to measurements in atmosphere were found.
Figure2shows the results of photoconductivity measure-ments for nanowires with three different thicknesses d:500 nm,190nm,and70nm.Their behavior concerning photo-current decay after UV illumination is completely different. For relatively thick wires(d>200nm)in the dark,currents between10-6A and10-5A are measured in both voltage polarities.After illumination with UV light by a mercury-xenon lamp via a quartz fiber(approximately15W/cm2),a photocurrent higher by slightly an order of magnitude is observed.After switching off the light the photocurrent
*Corresponding author.E-mail:r.calarco@fz-juelich.de.
NANO LETTERS
2005 Vol.5,No.5 981-984
decays fast,but not to the dark current value;it rather stays as a persistent photocurrent(in the dark),even120s after the end of the illumination at approximately double the dark current value(Figure2a).This effect of a persistent photo-current is observed for wires with diameters down to about 100nm.For the example in Figure2b with d)190nm the photocurrent upon UV illumination decreases each time when the light is switched off for some short period?t)1-2s, but does not reach the dark current value.A considerable current persists in the dark periods.The effect is also clearly seen from the time transient of the photocurrent in the500 nm sample(Figure3),where even after5min the current exceeds the initial dark value by about3×10-6A.
Thinner wires with diameters below about100nm(see example in Figure2b with d)70nm),in contrast,exhibit a fast photoresponse.For the70nm sample the maximum observed photocurrents reach values at around10-8 A. Control experiments have shown that the measured dark current of this sample is a parasitic background current not related to transport through the GaN column.Photocurrents in these thin GaN columns decay rapidly in the dark,which demonstrates that there is no persistent photocurrent anymore. Such a fast photoresponse was also observed by Han et al.18 on15nm thin GaN nanowires,but without an interpretation of the effect.
This general pattern of size dependent photocurrents is also clearly seen when one plots the measured photocurrent (as additive to dark current or parasitic background),at a bias voltage of about1V;more precisely,an electric field of1V/μm,to account for different electrode spacings,versus GaN nanowire diameter(Figure4).While for wire diameters above100nm the photocurrent changes only slightly with diameter,below approximately80nm a sharp drop to immeasurable values is observed.
Figure1.SEM picture of a nanowire on Si host substrate,with
Ti/Au contact electrodes.Nanowire diameter:70nm.
Figure2.Current-voltage characteristics of GaN nanowires with different diameters with and without UV illumination.(a)500nm sample,dark and under steady-state UV illumination.(b)190nm and70nm samples,dark and under steady-state UV illumination, as well as under periodic UV illumination(dash-dotted).The behavior of the current after switching off the light(persistent photoconductivity)depends on the diameter.Figure3.Current transient after1min of UV illumination(15 W/cm2)of the500nm sample.
Figure4.Photocurrent with UV illumination of approximately 15W/cm2versus whisker diameter.The kink in the fitting curve at85nm indicates the critical diameter d crit,where the surface depletion layer just completely depletes the nanowire.For smaller diameters the photocurrent shows an exponential decrease,for larger diameters the photocurrent is proportional to the wire diameter.
A double logarithmic plot of the current-voltage char-acteristics(Figure5)is helpful for a better understanding of the phenomena.For the190nm sample,the photocurrent under stationary illumination follows essentially a linear dependence on bias voltage,as is expected for ohmic behavior of a conducting channel in the whisker.Further-more,this ohmic behavior,which is also found for the500 nm thick sample in Figure2a,both in the dark and with UV illumination(at least up to0.4V),clearly demonstrates that the contacts are largely ohmic and that effects due to Schottky barriers at the contacts can be neglected.This is expected from literature data for Ti/Au contacts.19However, the dark current,3orders of magnitude below the photo-current at bias voltages between0.01and0.1V,turns into a V2to V3voltage dependence at bias voltages above0.1 V.This behavior is characteristic for space charge limited currents in insulators.20A similar effect is also observed for the90nm column.The photocurrent is essentially linear (ohmic)in voltage,while the dark current is below the measurement limit for voltages below1V and then shows a space charge limited current behavior.
From the observation of space charge limited currents we conclude that dark and photocurrents in these GaN columns are governed by depletion space charge layers,which are due to Fermi-level pinning at the surface of the nanocolumns. From previous photoemission spectroscopy studies on GaN surfaces,Fermi-level pinning is expected at about0.5-0.6 eV below the conduction band edge.21Assuming an n-type background doping in the1017cm-3range,the depletion space charge layers should have extensions of50to100nm into the bulk.Depending on the column diameter the whiskers, therefore,are completely depleted at small diameters(<80 nm)or have a tight open conducting channel at diameters above100nm.But even under those conditions the columns are essentially insulators in the dark due to their depletion space charge layers.Dark currents,at least for diameters below200nm,are space charge limited,as is characteristic for insulators.Only the higher carrier densities under UV light illumination allow ohmic behavior for diameters above 100nm.
The existence of depletion space charge layers at the column surface also explains the size dependent photocurrent behavior(Figs.2,4)quantitatively.
Because of Fermi-level pinning at the surface,the elec-tronic bands,conduction(E C)and valence(E V),are bent upward at the surface of the n-doped column as shown schematically in Figure6.Electrons prefer the inner part of the column,whereas holes tend to move to the surface.Due to their spatial separation,recombination of nonequilibrium carriers is thus reduced or is maybe even impossible if recombination via surface traps in the forbidden band is the prevailing recombination mechanism.Electrons would have to surpass the conduction band barrier at the surface for surface recombination.This model of hindered surface recombination due to the presence of depletion space charge layers explains the persistent photocurrent for columns with diameters above100nm(Figures2and3).As is seen qualitatively from Figure6,a decrease of the column diameter leads to complete depletion(detail in the middle) at a critical diameter d crit with unchanged surface barrier heightΦfor electrons in the conduction band.Down to this critical diameter,the recombination rate,i.e.,also the photocurrent,does not change significantly.Further shrinking of the dimensions,however,causes less band curvature and therefore a reduction of the barrier for surface electron-hole pair recombination(Figure6,detail on the left).With decreasing thickness now the recombination process is strongly enhanced and the photocurrent decays strongly with decreasing barrier height,i.e.,with decreasing column thickness.
This model thus explains that the photocurrent remains on a relatively high level as long as the column diameters exceed the critical value around80to100nm,where the column is completely depleted.Below this critical diameter surface barriers are diminished and the photocurrent drops steadily due to enhanced surface recombination(Figure6). The following assumptions allow a quantitative model description for the observed size dependent photoconductiv-
Figure5.Double logarithmic plot of the current-voltage char-acteristic for samples with different diameters.For comparison I∝V and I∝V2dependences are given in dash-dotted line.The photo-current(UV)shows ohmic behavior,while the dark current turns into a V2to V3voltage dependence at bias voltages above0.1V. This behavior is characteristic for space charge limited currents in insulators.(The dark current of the thin nanowire is below the measurement limit for voltages below1V).Figure6.Dependence of depletion region(shaded),shape of conduction(E C)and valence band edges(E V),and recombination barrierΦon the nanowire diameter d.The relative energetic locations of E C,E V,and E F are not on scale.The detail on the right shows the surface recombination mechanism of the photoexcited carriers.
ity.(i)The nanocolumns behave like a photoconductor without Schottky-contacts at the metallization.(ii)The optical absorption length for photons is much larger than the column diameter,such that photoexcitation is homogeneous all over the sample.(iii)Photogenerated carriers do not significantly affect the potential,i.e.,the band structure within the column. (iv)The time decay of the persistent photocurrent is essentially due to surface recombination,where holes are pushed to the surface due to the surface band bending and electrons have to overcome the corresponding surface barrier (Figure6)in order to recombine.Under these conditions we calculate the stationary photocurrent from the balance between generation rate(proportional to absorption constant, column volume,and irradiated light intensity)and surface recombination rate(proportional to column surface and recombination rate).The recombination rate is essentially given by an exponential term exp(-φ/kT),where the barrier for electrons to recombine with holes amounts to
For d>d crit(minimum diameter for a fully depleted column) a constant barrier height(eq1b)with N D as donor concentra-tion and 0as dielectric constant of GaN causes a nearly constant(slightly varying)photocurrent(proportional to column diameter),as is observed in the experiment(Figure 4).For column diameters d In conclusion,our model of surface electron-hole pair recombination excellently describes the unusual behavior of size dependent persistent photocurrents in GaN nanocolumns. The physical bases of the effect are size dependent recom-bination barriers within the whiskers,which are due to the interplay between column diameter and space charge exten-sion. The described effects are of general importance for all kinds of semiconductor nanowires,electronic transport through these structures,and in particular applications of nanowires in optoelectronics and sensor technology. References (1)Lauhon,L.;Gudiksen,M.;Wang,D.;Lieber,C.Nature2002,420, 57. (2)Gudiksen,M.;Lauhon,L.;Wang,J.;Smith,D.;Lieber,C.Nature 2002,415,617. (3)Wang,J.;Gudiksen,M.;Duan,X.;Cui,Y.;Lieber,C.Science2001, 293,1455. (4)Hiruma,K.;Yazawa,M.;Haraguchi,K.;Ogawa,K.;Katsuyama, T.;Koguchi,M.;Kakibayashi,H.J.Appl.Phys.1993,74,3162. (5)Ohlsson,B.;Bjo¨rk,M.;Magnusson,M.;Deppert,K.;Samuelson, L.;Wallenberg,L.Appl.Phys.Lett.2001,79,3335. (6)Ohlsson,B.;Bjo¨rk,M.;Persson,A.;Thelander,C.;Wallenberg,L.; Magnusson,M.;Deppert,K.;Samuelson,L.Physica E2002,13, 1126. (7)Sa′nchez-Pa′ramo,J.;Calleja,J.;Sa′nchez-Garcia,M.;Calleja,E.;Jahn, U.Physica E2002,13,1070. (8)Calleja,E.;Sa′nchez-Garcia,M.;Sa′nchez,F.;Calle,F.;Naranjo,F.; Mun?oz,E.;Jahn,U.;Ploog,K.Phys.Re V.B2000,62,16826. (9)Ristic,J.;Calleja,E.;Sa′nchez-Garcia,M.;Ulloa,J.;Sa′nchez-Pa′ramo, J.;Calleja,J.;Jahn,U.;Trampert,A.;Ploog,K.Phys.Re V.B2003, 68,125305. (10)Thelander,C.;M?rtensson,T.;Bjo¨rk,M.;Ohlsson,B.;Larsson,M.; Wallenberg,L.;Samuelson,L.Appl.Phys.Lett.2003,83,2052. (11)Greytak,A.;Lauhon,L.;Gudiksen,M.;Lieber,C.Appl.Phys.Lett. 2004,84,4176. (12)Duan,X.;Huang,Y.;Agarwal,R.;Lieber,C.Nature2003,421, 241. (13)Zhong,Z.;Wang,D.;Cui,Y.;Bockrath,M.;Lieber,C.Science2003, 302,1377. (14)Bjo¨rk,M.;Ohlsson,B.;Thelander,C.;Persson,A.;Deppert,K.; Wallenberg,L.;Samuelson,L.Appl.Phys.Lett.2002,81,4458. (15)Bjo¨rk,M.;Ohlsson, B.;Sass,T.;Persson, A.;Thelander, C.; Magnusson,M.;Deppert,K.;Wallenberg,L.;Samuelson,L.Appl. Phys.Lett.2002,80,1058. (16)Lu¨th,H.Solid Surfaces,Interfaces and Thin Films,4th ed.; Springer:Berlin,Heidelberg,2001;Chapter7. (17)Calarco,R.;Marso,M.;Meijers,R.;Richter,T.;Aykanat,A.I.; Stoica,T.;Lu¨th,H.Proceedings of the ASDAM‘04Conference, Smolenice,Slovakia,2004,9. (18)Han,S.;Jin,W.;Zhang,D.;Tang,T.;Li,C.;Liu,X.;Liu,Z.;Lei, B.;Zhou, C.Chem.Phys.Lett.2004,389,176. (19)Schmitz,A.C.;Ping,A.T.;Khan,M.A.;Chen,Q.;Yang,J.W.; Adesida,I.Semicond.Sci.Technol.1996,11,1464. (20)Rose,A.Phys.Re V.1955,97,1538. (21)Kocˇan,M.;Rizzi,A.;Lu¨th,H.;Keller,S.;Mishra,U.K.Phys.Status Solidi B2002,234,773. NL0500306 φ)eN D d2/16 for d crit (1a) φ)eN D d2 crit /16 for d>d crit (1b)